Machine Learning algorithms can be used to detect operational abnormalities, malicious activities, and security threats. Adversaries on the other hand try to evade or even poison and subvert these algorithms. This requires designing machine learning algorithms that are resistant to ever-more-resourceful adversaries. In this talk, we will provide an overview of the interplay between attack and defense in design of secure machine learning algorithms. We will discuss attacks against these algorithms such as exploratory attacks, causative attacks, and reverse engineering. We will also present techniques to design and develop machine learning algorithms having the required robustness for adversarial environments, and provide clarifying examples to show how to apply these techniques.
Bahman did his PhD at Stanford University, supported by William R. Hewlett Stanford Graduate Fellowship, and focused on the topic of algorithms for big data applications, in which he is a well-published author in some of the best conferences and journals, including PVLDB, SIGMOD, WWW, and KDD. He was the last PhD student of the legendary late Rajeev Motwani, and has been also advised and co-advised by Ashish Goel and Prabhakar Raghavan (formerly Yahoo VP of Strategy, currently Google VP of Engineering). His industry experience during his PhD studies spans several internships and collaborations with some of the best researchers and practitioners from Twitter, Microsoft Research, Yahoo Research, AOL, and Google. He is a recipient of the Yahoo Key Scientific Challenges Award for his contributions to the area of search technologies.
Comments on this page are now closed.